CN104795076B - A kind of audio method for detecting - Google Patents

Abstract

The invention discloses an audio detection method, comprising the following steps: after the audio monitoring is turned on, sampling the environmental background noise and calculating the energy value corresponding to each sampling point, and comparing all the energy values in the sampling time of 20 to 40 seconds The energy value corresponding to the sampling point is obtained to obtain the maximum value and the minimum value; then the sound in the monitoring area is sampled with a sampling time interval Δt of 1/64 to 1/2 second and the energy value corresponding to each sampling point is calculated. The energy value corresponding to each sampling point obtained in the time length S2 is compared with the alarm reference threshold, and the alarm is triggered when the number of sampling points greater than the alarm reference threshold exceeds the value N/2, and the number of sampling points greater than the alarm reference threshold is less than An alarm will not be triggered when the value is N/2, and N is the number of sampling points S2/Δt within the sampling time length S2. The method can exclude regular noise or sudden noise or interference in the surroundings to detect abnormal sounds.

Description

A method of audio detection

technical field

The invention belongs to the field of monitoring equipment, in particular to an audio detection method.

Background technique

For pure video monitoring, the effect is not good at night and in cloudy conditions, while the sound signal is not affected by light and weather. Audio monitoring products include pickups, players, voice mail systems, recording cards, etc., and the core is the pickup. For audio and video integrated monitoring, the camera is the eyes, and the pickup is the ear. However, it is purely based on the sound level to detect whether to alarm, such as thunder, car whistle, firecrackers and other natural sounds will also trigger the alarm, causing false alarms. At the same time, it cannot be used in noisy environments, and it is very easy to miss and false positives.

Contents of the invention

In order to overcome the deficiencies of the prior art, the present invention provides an audio detection method, which uses audio to determine whether there is an illegal intrusion, and detects abnormal sounds. This method can eliminate the presence of regular noise or sudden noise around Influence or interference, the alarm will not be triggered by the natural sound of a nearby thunder, a car whistle, etc., and it will not lose its effect near the downtown area with complex background noise, so this method is mainly used for more complex Sound detection of security monitoring system in environment.

The technical scheme provided by the invention is:

A method for audio detection, comprising the steps of:

Step 1. After the audio monitoring is turned on, sample the environmental background noise in the monitoring area at a sampling rate of 8k-44.1kHz and calculate the energy value corresponding to each sampling point. The sampling time length S1 is 20-40 seconds. By comparing The energy values corresponding to all sampling points in this time period are obtained to obtain the maximum value E _max and the minimum value E _min ;

Step 2. After sampling the ambient background noise in the monitoring area for 20 to 40 seconds as described in step 1, then sample the sound of the monitoring area with a sampling time interval Δt of 1/64 to 1/2 second and calculate The energy value corresponding to each sampling point, the sampling time length S2 is the time length of the abnormal sound expected to be monitored by the user, which is 5 to 60 seconds, and the energy value corresponding to each sampling point obtained within the sampling time length S2 is equal to Compared with an alarm reference threshold, the calculation method of the alarm reference threshold is E _max +(E _max -E _min )*x, wherein, x is the noise sensitivity set by the user, if it is greater than the sampling of the alarm reference threshold When the number of points exceeds the value N/2, an alarm is triggered. If the number of sampling points greater than the alarm reference threshold is less than the value N/2, the alarm is not triggered. N is the number of sampling points in the sampling time length S2 S2/Δt;

Step 3, when the number of sampling points greater than the alarm reference threshold in the sampling time length S2 described in step 2 is less than the value N/2, repeat the steps in step 2;

Step 4: Repeat steps 1, 2 and 3 every 10-60 minutes.

Preferably, in the audio detection method, in step 1, the ambient background noise in the monitoring area is sampled at a sampling rate of 44.1 kHz, and the sampling time length S1 is 30 seconds.

Preferably, in the audio detection method, the sampling time interval Δt in step 2 is 1/16.

Preferably, in the audio detection method, the value of noise sensitivity x set by the user in step 2 is 0.1-10.0.

Preferably, in the audio detection method described in step 2, after the number of sampling points greater than the alarm reference threshold within the sampling time length S2 in step 2 is greater than the value N/2 to trigger the alarm, repeat steps 1 and 2 Implementation process.

Wherein, when calculating the energy value corresponding to the sampling point in the sampling time length S1 or the sampling time length S2 described in step 1 and step 2, the energy value is obtained by accumulating the sound wave area in an integral manner.

Compared with the prior art, the audio detection method of the present invention can operate in a relatively constant noise environment to prevent missing alarms. The method can eliminate regular noises around, such as air conditioners, motors, etc. Such as the sound of long-term operation, the noise near the downtown area, the sound of cars, the sound of machine tools in the factory environment, etc. The second is to eliminate the interference of natural factors such as thunder, car whistle, etc., and prevent false alarms. The third is to use the integration method to detect abnormal sounds, which is more reliable than simple comparison thresholds.

Description of drawings

FIG. 1 is a flow chart of the audio detection method of the present invention.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings, so that those skilled in the art can implement it with reference to the description.

As shown in Figure 1, the present invention provides an audio detection method, comprising:

Step 1. After the audio monitoring is turned on, sample the environmental background noise in the monitoring area with a sampling rate of 44.1kHz and a sampling depth of 8/16bit and calculate the energy value corresponding to each sampling point. The sampling time length S1 is 30 seconds. When calculating the energy value corresponding to the sampling point, the integral method is used to accumulate the sound wave area to obtain the energy value of the sound. In this way, a set of energy values is obtained. By comparing the energy values in the group, the maximum value E _max and the minimum value E _min are obtained. ;

Step 2. After sampling the ambient background noise in the monitoring area for 30 seconds as described in step 1, then sample the sound of the monitoring area with a sampling time interval Δt of 1/16 second and calculate the corresponding Energy value, the sampling time length S2 is the time length of the abnormal sound expected to be monitored by the user, which is 5 to 60 seconds, and the energy value corresponding to each sampling point obtained within the sampling time length S2 is compared with an alarm reference threshold In comparison, the calculation method of the alarm reference threshold is E _max +(E _max -E _min )*x, where x is the noise sensitivity set by the user, if the number of sampling points greater than the alarm reference threshold exceeds the value N /2, then trigger an alarm, if the number of sampling points greater than the alarm reference threshold is less than the value N/2, then do not trigger an alarm, N is the number of sampling points S2/Δt in the sampling time length S2;

Step 3, when the number of sampling points greater than the alarm reference threshold in the sampling time length S2 described in step 2 is less than the value N/2, repeat the steps in step 2;

Step 4, E _max and E _min alarm reference thresholds can be updated every 10-60 minutes according to the actual needs of users, in order to adapt to changes in the background noise environment and prevent false alarms and false alarms, so every 10-60 minutes Minutes, repeat the execution process of step 1, step 2 and step 3.

In the audio detection method, the value of noise sensitivity x set by the user in step 2 is 0.1-10.0.

In the audio detection method described in step 2, after the number of sampling points greater than the alarm reference threshold within the sampling time length S2 in step 2 is greater than the value N/2 to trigger an alarm, the execution process in step 1 and step 2 is repeated.

Although the embodiment of the present invention has been disclosed as above, it is not limited to the use listed in the specification and implementation, it can be applied to various fields suitable for the present invention, and it can be easily understood by those skilled in the art Therefore, the invention is not limited to the specific details and examples shown and described herein without departing from the general concept defined by the claims and their equivalents.

Claims (5)

1. a kind of audio method for detecting, which is characterized in that include the following steps：

Step 1: after Voice Surveillance is opened, the environmental background noise of monitoring area is adopted with the sample rate of 8k~44.1kHz Sample simultaneously calculates the corresponding energy value of each sampled point, and sampling time length S1 is 20~40 seconds, by comparing the period In the corresponding energy value of all sampled points, obtain maximum value E_maxWith minimum value E_min；

Step 2: after the environmental background noise to monitoring area described in step 1 carries out sampling in 20~40 seconds, then The sound of monitoring area is sampled with 1/64~1/2 second sampling time interval Δ t and calculates each sampled point pair The energy value answered, sampling time length S2 are time span 5~60 existing for the abnormal sound set by user for expecting to listen to Second, the corresponding energy value of each sampled point obtained in sampling time length S2 is compared with an alarm reference threshold Compared with the computational methods of the alarm reference threshold are E_max+(E_max-E_min) * x, wherein x is the noise sensitivity of user setting, When being more than numerical value of N/2 if it is greater than the sampled point number of the alarm reference threshold, then alarm is triggered, if it is greater than the alarm When the sampled point number of reference threshold is less than numerical value of N/2, then alarm is not triggered, N is that the sampled point in sampling time length S2 is a Number S2/ Δs t；

Step 3: the sampled point number in the S2 of sampling time length described in step 2 more than the alarm reference threshold is less than number When value N/2, the step in step 2 is repeated；

Step 4: every 10-60 minutes, the implementation procedure Step 1: step 2 and step 3 is repeated.

2. audio method for detecting as described in claim 1, which is characterized in that in step 1, with the sample rate of 44.1kHz to prison The environmental background noise in control region is sampled, and sampling time length S1 is 30 seconds.

3. audio method for detecting as claimed in claim 2, which is characterized in that the sampling time interval Δ t described in step 2 It is 1/16.

4. audio method for detecting as claimed in claim 3, which is characterized in that the noise of the user setting described in step 2 is quick Sensitivity x numerical value is 0.1~10.0.

5. audio method for detecting as described in claim 1, which is characterized in that big in the S2 of sampling time length described in step 2 After the sampled point number of the alarm reference threshold triggers alarm more than numerical value of N/2, holding in step 1 and step 2 is repeated Row process.